ande 0.0.1

Table of Contents

1. About The Project
   • Built With
2. Getting Started
   • Prerequisites
   • Installation
3. Usage
4. Roadmap
5. Contributing
6. License
7. Contact
8. Acknowledgments

About The Project

The world of data is growing very fast, and it is a new challenge for data analysis to develop new methods to handle this massive amount of data. A large number of data have many hidden factors that need to be identified and used for different algorithms. Clustering is one of the significant parts of data mining. The term comes from the idea of classifying unsupervised data. Now-a-days a lot of algorithms are implemented. Besides that, all those algorithms have some limitations, creating an opportunity to innovate new
algorithms for clustering. The clustering process can be separated in six different ways: partitioning, hierarchical, density, gridmodel, and constraint-based models. The aim of the package is to implement various types of clustering algorithms and helps to determine which one is more accurate on detecting impure data from a large data set. To create a common platform for Some popluar algorithms for anomaly detection are implemented and converged all of them into a package(AnDe). The algorithms which are implemented and combined into the package are: K-means, DBSCAN, HDBSCAN, Isolation Forest, Local Outlier Factor and Agglomerative Hierarchical Clustering. The package reduce the consumption of time by compressing implementation hurdles of each algorithms. The package is also makes the anomaly detection procedure more robust by visualizing in a more precise way along with visualization of comparison in performance(accuracy, runtime and memory consumption) of those algorithm
implemented.

Built With

For using this package, some popular packages are need to be configured in the working environment.

• Numpy
• Panda
• Matplotlib
• Time
• os
• Sklearn
• Hdbscan
• Tracemalloc

(back to top)

Getting Started

This is an example of how you set up thie pacage and use in you script.

Prerequisites

At first, need install the package in your working environment for using this package.

pip install python=3.8

pip install numpy

pip install pandas

pip install matplotlib

pip install time

pip install os

pip install sklearn

  pip install Hdbscan

pip install Tracemalloc

Installation

1. Download the package from (https://github.com/cbiswascse/AUnifiedPackageForAnomalyDetection)
2. Install the package in you environment.

   pip install cb-cluster
   

3. Import the pacage in your script.

   from EMOADCAUP import Cluster
   

(back to top)

Usage

1. Call the cluster function.

   from ande import ande 
   ande.ClusterView()
   

2. Input the Location of CSV file.

   Please, Input the Location of CSV:
   

3. Select yes(y) If you have Catagorical data in your dataset.

   Do you want to include Catagorical data [y/n]:
   

4. Select yes(y) If you want to scaling your dataset with MinMaxScaler.

   Scaling data with MinMaxScaler [y/n]:
   

5. Available Clusering Algorithm Kmeans Dbscan Isolation Forest Local Factor Outlier Hdbscan Agglomerative

   Choose your Algorithm:
   

6. Kmeans Clusering: Number of Cluster

   How many clusters you want?:

7. Select one of Average Method for Performance Metrics

   weighted,micro,macro,binary

8. Dbscan: Input a Epsilon value

   epsilon in Decimal:
   

9. Input a Min Samples value

   Min Samples In Integer:

10. Select one of Average Method for Performance Metrics

 weighted,micro,macro,binary

11.Hdbscan: Minimum size of cluster

	Minimun size of clusters you want?:

12. Select one of Average Method for Performance Metrics

   weighted,micro,macro,binary

13.Isolation Forest: Contamination value

   Contamination value between [0,0.5]:

14. Select one of Average Method for Performance Metrics

   weighted,micro,macro,binary

15. Local Outlier Factor: Contamination value

   Contamination value between [0,0.5]:

16. Select one of Average Method for Performance Metrics

   weighted,micro,macro,binary

17.Agglomerative: Number of Cluster

   How many clusters you want?:

18.Select one of Average Method for Performance Metrics

   weighted,micro,macro,binary

(back to top)

License

Distributed under the MIT License. See LICENSE.txt for more information.

(back to top)

Contact

Chandrima Biswas - cbiswascse26@gmail.com

Project Link: https://github.com/cbiswascse/AUnifiedPackageForAnomalyDetection

(back to top)

Acknowledgments

I would like to convey my heartfelt appreciation to my supervisor Prof.Dr. Doina Logofatu,for all her feedback, guidance, and evaluations during the work. Without her unique ideas, as well as her unwavering support and encouragement, I would never have been able to complete this project. In spite of her hectic schedule, she listened to my problem and gavethe appropriate advice.
Furthermore, I express my very profound gratitude Prof. Dr. Peter Nauth for being the second supervisor of this work.